As transportation agencies adopt artificial intelligence (AI) tools to improve and expedite their work and to improve transportation systems, it’s worth also exploring how AI impacts transportation policy through its own land use needs. Data centers, which are the physical infrastructure behind AI software, consume significant amounts of electricity, land, and water in their operations. Data centers have been around since the start of the internet age. However, the demands of the AI boom have resulted in the rapid development of hyperscale data centers around the United States. For transportation agencies and policymakers, impacts of data centers include competition in the transportation construction market, demand to the electrical grid, and the need to strike a balance between building infrastructure more efficiently and addressing localized concerns.

The rapid built out of data centers has resulted in increased public awareness of the physical impacts of AI and increased policy debate at local, state, and federal levels on the existing regulations and incentives for data center build out. Low public opinion of data centers has been capitalized on by candidates during the midterm election cycle across party lines. As the conversation around AI continues, transportation policymakers must consider how AI will impact transportation both as tool and as development sites.

Transportation Construction Impacts

The deployment of data centers has direct transportation impacts. Developers may be required to complete traffic impact analyses in the permitting process, and the development of data centers on unused lands may require increased roadway build out. Increased traffic from data centers primarily occurs during the construction phase. If replacement of servers is required periodically throughout the life of the data center, then there will also be potentially increased traffic during these periods. However, once the data center is built, their traffic impacts are limited to the daily traffic of workers at the data center. Large data centers can employ around 25 to 150 permanent workers including technicians and security staff. Housing, retail, or other forms of commercial real estate of similar size would likely generate much more significant traffic.

In addition to the direct roadway impacts, increased data center build out impacts roadway and highway construction due to increased competition for materials and labor. According to the US Census Bureau, data center construction accounts for 2.3% of US construction spending as of June 2026. From June 2025 to June 2026, private sector spending on data center construction increased 28% while private spending on total non-residential construction decreased 4%. Increased competition from the private sector for construction labor and materials may result in increased prices or wait times for public sector transportation infrastructure projects.

North America’s Building Trades Unions has been a proponent of data center build out due to construction job creation. The AFL-CIO has called for data centers to be constructed with union labor.

Energy Usage and a Constrained Electrical Grid

Between 2018 and 2023, the electricity consumption of data centers increased from 1.9% of the United States’ total consumption to 4.4% and is projected to potentially hit 12% by 2028. The increase in data center build out comes at a time when the U.S. electric grid is already constrained. The build out of additional generation, transmission, and distribution infrastructure to meet the needs of data centers could increase electric utility costs which could then be passed on to ratepayers.

The Trump Administration’s calls on hyperscale data centers to cover the full additive cost to energy infrastructure, negotiate separate rate structures, and coordinate with grid operators to provide backup generation. While major companies have publicly agreed to the pledge, action will be needed by states and utilities to ensure its aims materialize. For example, in 2025, Texas enacted Senate Bill SB6 which directs the Public Utility Commission of Texas to determine a reasonable payment sharing structure for upgrade costs of large loads. The Public Utility Commission’s draft final rule (published March 2026), includes an upfront interconnection fee and interconnection study fees.

Data center development is running into a similar issue as industries and advocates pushing to electrify transportation: conflict between increasing demand for electricity and an aging electrical grid. Demand from data centers on the electric grid is more intensive and less variable than demand from electric vehicle charging infrastructure. However, policy questions around who should pay for increased demand for electricity usage are relevant to both electrified transportation and data centers. The question of who pays for such upgrades and the speed at which they occur remains an open question.

Land Use and Permitting Debates

In addition to the impact of data centers on the electrical grid, across the country, states and localities are considering the wider land use impacts of data centers. Developments are relatively concentrated; 60 percent of the nation’s data centers are located in just ten states. Virginia and Texas lead the pack in data center development. Given the increased scale of new data center developments, many companies are selecting rural locations with more open space. Environmental advocates have raised concerns about land use selections in areas with already constrained water and energy resources.

Location selection decisions are also motivated by the permitting requirements and incentives provided by different states and localities. These can take the shape of tax incentives, permitting requirements, zoning classifications, and potential energy cost sharing requirements. As public awareness and organized opposition to data centers has grown, states and localities have considered or moved forward measures to increase friction in the permitting process. At the state level, legislation has been introduced to slow or ban data center construction by several state legislatures. In July, the first statewide moratorium was signed by New York Governor Katy Hochul, with the expressed intention of giving the state time to create guidelines to protect ratepayers and the environment.

Even in states that have historically been supportive of data center developments, there has been political and public pushback to the current wave of AI-investment fueled construction. Since the start of the internet age, Virginia has served as an epicenter for data centers and internet traffic movement due to policy incentives and extensive fiber networks. At the state level, Virginia provides a data-center sales-tax exemption. Virginia legislators have mixed views on new data center development. In 2025, the legislature passed a bill to establish statewide land use reviews for data center construction but it was vetoed by then-Governor Glen Youngkin. However, at the local level, counties in Virginia have moved forward with actions to change the permitting processes for data centers. In Fairfax County, Viriginia, zoning was updated to create a separate use for data centers. Loudoun County, Virigina used to have by-right approval for data centers but now requires an exception process with public hearings. From May 2024 to March 2025, opposition to data centers resulted in delays or cancellations of around $64 billion worth of projects.

Most states continue to offer tax incentives for data centers, seeking to attract deployments in their states, highlighting the potential value that states and localities see in hosting data centers. It is estimated that Virginia’s data center sales tax exemption cost the state $1.6 billion in potential revenue in FY2025. At the same time, real and property tax revenues collected from data centers are significant. In Loudon County, in FY2025 $900 million was collected from data centers. Data centers are estimated to have raised $162 billion in government revenues in 2023.

Concurrently, the federal government has pushed for streamlined federal permitting of data centers. In July 2025, President Trump signed Executive Order 14318, calling for expedited permitting through modifications to the Clean Air Act, the Clean Water Act, the Toxic Substances Control Act, and other relevant federal regulations.

Across infrastructure types, there have been both Administration efforts and bipartisan efforts in Congress to expedite federal permitting processes. The federal push for expedited data center permitting contrasts with the increased resistance to data centers emerging at the local level. This provides an important lesson for federal policymakers: efforts to expedite permitting at the federal level will continue to intersect with local concerns. For developers, community engagement and buy in will continue to play a role even with support from the federal government.

Data centers also run into a related policy concern that commonly emerges in infrastructure development: while the externalities of a deployment are concentrated locally and regionally, the outputs of the infrastructure are dispersed at a larger scale. Unlike certain forms of transportation infrastructure like transit, living near a data center does not inherently provide the from the infrastructure aside from tax revenues and somewhat limited job creation. Because of this (and because the internet and artificial intelligence are not public goods in the same way transportation infrastructure is), there is increased impetus on developers to make a case for the benefits of the infrastructure at both a local and national scale. Additionally, since data centers are not a public utility like transportation infrastructure, developers do not have any kind of eminent domain power, giving local governments and powers increased leverage.

Data center developments may at first glance seem to exist outside of transportation policy. However, data center build out has direct impacts on transportation construction markets. Data center developers must also contend with existing permitting processes and a restrained electrical grid that transportation infrastructure developers face. As data center developers navigate relationships with local counties, they can learn from community benefit agreements developed in other infrastructure spheres.

Federal policymakers and developers interested in building infrastructure more efficiently must take seriously the current wave of local pushback to data centers and find constructive ways to address localized needs across infrastructure types. The physical impacts of data centers including land, energy, and water use, must also be considered as transportation systems integrate artificial intelligence. While AI products can improve movement and connectivity, their own infrastructure imprint must be assessed in the larger conversation about the impacts of artificial intelligence on transportation.